Mitochondrial Reprogramming Underlies Resistance to BCL-2 Inhibition in Lymphoid Malignancies

Abstract

Mitochondrial apoptosis can be effectively targeted in lymphoid malignancies with the FDA-approved B cell lymphoma 2 (BCL-2) inhibitor venetoclax, but resistance to this agent is emerging. We show that venetoclax resistance in chronic lymphocytic leukemia is associated with complex clonal shifts. To identify determinants of resistance, we conducted parallel genome-scale screens of the BCL-2-driven OCI-Ly1 lymphoma cell line after venetoclax exposure along with integrated expression profiling and functional characterization of drug-resistant and engineered cell lines. We identified regulators of lymphoid transcription and cellular energy metabolism as drivers of venetoclax resistance in addition to the known involvement by BCL-2 family members, which were confirmed in patient samples. Our data support the implementation of combinatorial therapy with metabolic modulators to address venetoclax resistance.

Keywords: AMPK; BCL-2; CRISPR/Cas9; chronic lymphocytic leukemia; clonal evolution; drug resistance; genome-wide screen; metabolism; mitochondrion; venetoclax.

Conflict of interest statement

DECLARATION OF INTERESTS

C.J.W. is a co-founder of Neon Therapeutics, Inc and is a member of its scientific advisory board, and receives research funding from Pharmacyclics. R.G., Abbvie (honoraria, travel funds), Janssen (honoraria, travel funds), Gilead (honoraria, travel fund) and Roche (travel funds). J.R.B. serves as a consultant for Abbvie, Genentech, Astra-Zaneca, Janssen, Pharmacyclics, Gilead, Verastem, TG Therapeutics, Sunesis and Loxo and receives research funding from Gilead, Verastem and Sun. J.W.H. is co-founder of Rheostat Therapeutics and is member of its scientific advisory board and is also a member the scientific advisory board of X-Chem, Inc (honoraria). A.G.L. discloses consulting and laboratory research support from AbbVie, Novartis, and Astra-Zeneca; he is an equity-holding co-founder of Flash Therapeutics and Vivid Biosciences. V.K.M. is on the Scientific Advisory Board of Janssen Pharmaceuticals, is a venture partner consultant to 5AM Ventures, and a founder and equity holder in Raze Therapeutics. R.G. and C.J.W. disclose a patent related to this work ((U.S. Provisional Application No. 62/744,081). All other authors declare no competing interest.

Copyright © 2019 Elsevier Inc. All rights reserved.

Figures

Figure 1.. CLL cells from patients developing resistance to venetoclax undergo clonal evolution and exhibit complex trajectories.

(A) Schema of the 6 studied patients (Pt) with timing and sites of pre- and post-treatment sample collections indicated. (B) Comparison of the shifts in cancer cell fraction (CCF) in pre-treatment and relapse samples, demonstrating clonal evolution and diverse changes in subclonal composition across the 6 patients. Driver mutations associated with each clone are indicated. Superscripted numbers indicate distinct mutations of the same gene per patient. Del, deletion; amp, amplification. (C) Comparison (modal CCF with 95% confidence interval) between pre-treatment and relapse samples for select drivers previously reported as recurrently observed in CLL. Mut, mutations. See also Figure S1 and Tables S1-S4.

Figure 2.. BCL-2 family members, lymphoid transcription regulators and components of AMP-dependent pathways are candidate drivers of venetoclax resistance.

(A) Experimental schema of the parallel knockout and overexpression screens using the BCL-2 driven OCI-Ly1 cell line (two biologically independent experiments per screen). (B) sgRNAs frequencies at different timepoints during the screens (two independent experiments shown), black bars are mean +/− s.d., p value is from two-sided t-test. (C) Scatter plots showing the average log2fold-change (LFC) for each gene in both duplicates of the loss-of-function screens (only genes with LFC > −1 are shown). Genes with a significant increase of sgRNAs representation (using the gene-ranking algorithm STARS, Broad Institute) are highlighted. (D) ORF frequencies at different timepoints during the screens (two independent experiments shown), black bars are mean +/− s.d., p value is from two-sided t-test. (E) Scatter plots showing the average log2fold-change (LFC) for each gene in both duplicates of the GOF screens (only genes with LFC > −1 are shown). Genes within the top-30 ORFs are highlighted. (F) Dose-response curves to venetoclax of 2 representative single-knockout OCI-Ly1 cells with related western-blots for quantification of the target protein. Data are mean +/− s.e.m. (G) Cumulative growth over time of each of the genetically perturbed OCI-Ly1 cells. See also Figure S2 and Tables S5 and S6.

Figure 3.. Expression changes related to acquisition of venetoclax resistance involves MCL-1 and cellular energy metabolism.

(A) Dose-response curve of the drug-resistant OCI-Ly1-R and the drug-sensitive OCI-Ly1-S cell lines. Data are mean +/− s.e.m. (B) Scatter plot reporting log2fold-change (LFC) of expression levels of transcripts (X-axis) and proteins (Y-axis) levels between OCI-Ly1-S and OCI-Ly1-R cells. Red dots – events with adjusted p value < 0.05 at the protein level. (C) Western-blot showing MCL-1, BCL-XL and BCL-2 protein levels in OCI-Ly1-S and OCI-Ly1-R cells. (D) Dose-response curves of OCI-Ly1-S to venetoclax and varying doses of the MCL-1 inhibitor S63845. Data are mean +/− s.e.m. (E) Combination index according to the fraction affected (left) and normalized isobologram (right), Chou-Talalay method. (F) Viability of the OCI-Ly1-R line 24 hours after exposure to venetoclax 100 nM, S63845 50 nM and both drugs (and DMSO as control), data are mean +/− s.e.m. from three biologically independent experiments; p value is from ANOVA test with adjustment for multiple comparisons. See also Figure S3.

Figure 4.. Increased oxidative phosphorylation drives resistance to BCL-2 inhibition.

(A) Selected gene set enrichment analysis plots based on differential RNA expression between OCI-Ly1-S and OCI-Ly1-R. (B) Oxygen consumption rate (OCR) as a function of time in the OCI-Ly1 and SU-DHL4 lines, with exposure to inhibitors of the electron transport chain and oxidative phosphorylation (OXPHOS) to derive bioenergetics parameters of mitochondrial respiration. (C) Quantification of the reactive oxygen species superoxide by flow cytometry in resistant vs parental B-cell lines. (D) Ratio of mitochondrial DNA (mtDNA) over nuclear DNA (nucDNA) in resistant vs parental B-cell lines. (E and F) Oxygen consumption rate (OCR) (E) and extracellular acidification rate (ECAR) (F) over time in the OCI-Ly1 and SU-DHL4 lines upon treatment by venetoclax, with or without prior zVAD treatment, or DMSO as control. (G) Relative oxygen consumption rate (OCR) and over time in the control OCI-Ly1 and BAX/BAK1 double knockout OCI-Ly1 cell lines upon the treatment by venetoclax or DMSO as control. Data are mean +/− s.e.m. from three biologically independent experiments (panel C) and one representative experiment of three biological replicates (panel B, D, E, F and G), * denotes p

Figure 5.. Targeting AMPK or the mitochondrial electron transport chain modulates sensitivity to venetoclax.

(A) Dose-response curves of OCI-Ly1, SU-DHL4, SU-DHL6 and Toledo cell lines to increasing doses of venetoclax alone or in combination with the AMPK inhibitor dorsomorphin, the inhibitor of electron transport chain complex 3 antimycin and the F1Fo-ATPase inhibitor oligomycin. (B) Heatmap showing the Excess Over Bliss value related to the indicated combinations - venetoclax 5, 10 and 50 nM for OCI-Ly1 and 10, 50, 100, 5000, 1000, 5000 nM for SU-DHL4, SU-DHL6 and Toledo, dorsomorphin 1 and 2 μM, oligomycin 10 and 100 nM, antimycin 10 and 100 nM. (C) Heatmap of the mean viability from duplicate experiments of primary tumor cells collected from 10 CLL patients after exposure to various drugs combinations, as indicated (venetoclax 1 nM, dorsomorphin 2 μM, oligomycin 10 nM, antimycin 10 nM). (D) Dose-response curves of indicated cell lines to increasing doses of venetoclax with and without exposure to the AMPK-activator A-769662. (E) Viability of the resistant OCI-Ly1-R and SU-DHL4-R cell lines after exposure to venetoclax 500 nM with and without dorsomorphin 1 μM, antimycin 10 nM or oligomycin 10 nM (DMSO as control). For (A, C-E) viabilities were assessed at 24h of drugs exposure. For (A, D and E) the data shown represent mean +/− s.e.m. p value is calculated using a two-sided t-test. (F) Tumor volume measurements of NSG OCI-Ly1 xenografts treated for 20 days, with vehicle control (black line), Oligomycin (200 mg/kg, i.p., pink line), Venetoclax (25 mg/kg, p.o., dashed light blue line), or their combination (dark blue line). Data are mean +/− s.d. p values result from Repeated-Measures ANOVA test with Tukey’s correction. See also Figure S4andTable S1.

Figure 6.. A circuit of ID3 repression and PKA-AMPK deregulation is implicated in venetoclax resistance.

(A) Heatmap of differentially expressed transcripts between the OCI-Ly1-S and OCI-Ly1-R cells, and of knockout (KO) cell lines from the screen hits vs cell lines with KO using non-targeting sgRNAs. Relevant genes affected in common with the OCI-Ly1-S and –R cells and in the ID3 KO line are indicated. (B) Volcano plot of transcripts changes in ID3 KO OCI-Ly1 cells compared to non-targeting sgRNA transduced OCI-Ly1 cells. (C) Volcano plot of enriched accessible transcription factor motifs comparing ID3 KO vs control. (D) Western-blot quantification of ID2 and ID3 proteins in PRKAR2B and PRKAA2 overexpressing OCI-Ly1 cell lines. (E) Schema of the ID3 and PRKAR2B resistance circuit. (F) Sensitivity of PRKAA2 and PRKAR2B overexpressing OCI-Ly1 cells to venetoclax when used in combination with dorsomorphin (2 μM) and oligomycin (1 μM), compared to DMSO control. Data are mean +/− s.e.m. from three biologically independent experiments, p value is from two-sided t-test. (G) Viability at 24 hours of single-cell clones from ID3 knockout OCI-Ly1 cells compared to non-targeting sgRNAs transduced OCI-Ly1 cells after exposure to dorsomorphin and oligomycin in addition to venetoclax. Data are mean +/− s.d. from three biologically independent experiments and p values are from ANOVA test. See also Figure S5.

Figure 7.. Deregulated MCL-1 and AMPK signaling detected in CLL samples from patients developing resistance to venetoclax.

(A) Comparison of somatic copy number variations in the OCI-Ly1-S and OCI-Ly1-R cells. The red oval indicates amplification 1q as the main difference between the two lines. Red-gain; blue - loss. (B) Comparison (modal CCF with 95%CI) between pre-treatment and relapse leukemia samples from patients 1, 2 and 3 for amp(1q). (C) Representation of the minimal gained region in the 1q locus across the OCI-Ly1-R cell line and relapsed samples from Patients 1, 2 and 3. (D) Immunohistochemical stains of patient samples before and after progression on venetoclax for MCL-1, AMPK, ACC, and p-ACC, with representative images from Patient C and 2 (left, the scale bar represents 20 μm in Patient 2 and 50 μm in Patient C), and quantification of % positively staining cells before (grey) and after (red) venetoclax treatment for Patients 2-6 (right). Data are mean +/− s.e.m. from replicates; p value is from Welch t-test. (E) Progression-free survival according to MCL-1 expression (low, or = 10%). p value is from log rank test. (F) Relative expression of ID3 by qPCR from patient sample RNA before (grey) and after (red) venetoclax treatment from Patients 2, 5, and 6; p value is from Welch t-test. The boxes extend from the 25th to 75th percentiles and the whiskers from min to max. (G) Proposed model for venetoclax resistance in lymphoid malignancies. See also Figure S6 and Tables S1-S4, andS8.